I am using Postgres + EF Core. I have a column called Name which I want to be unique. I have tried the following:
builder.HasIndex(s => s.Name).IsUnique();
but this allows "test123" and "TEST123" to be accepted. How can I add a case insensitive unique constraint with fluent api?
Or do I just need to create a NormalizedName column and add the unique constraint to that. Seems like a lot of work every time I want to add a unique constraint to a column.
If you're using EF Core 5.0, then see below (probably the better solution). Otherwise, I added a case insensitive check by making use of the citext extension as described here. There are some limitations, I won't list them here but you can read up on them in the previous link or directly on the PostgreSQL Docs.
Also ensure you have the postgres-contrib package installed to make use of this extension.
First, enable the extension by adding it to your model builder
modelBuilder.HasPostgresExtension("citext");
Then use the citext as the column type
builder.Property(s => s.Name)
.HasColumnType("citext")
EF Core 5.0
With EF Core 5.0 there's better support by making use of Collations. You can also check out the Npgsql docs in regards to PostrgeSQL. This gets over a bunch of limitations with citext above and gives you a lot more control.
So the (untested) steps are:
Create a collation as a database object using ICU. This will create a non-deterministic, case-insensitive ICU collation. If you need something else, check out the ICU docs.
modelBuilder.HasCollation("case_insensitive_collation", locale: "en-u-ks-primary", provider: "icu", deterministic: false);
Now on you column, you can add:
builder.HasIndex(s => s.Name)
.UseCollation("case_insensitive_collation")
.IsUnique();
If you achieved it some other way, I'd love to hear how.
Related
I use Code First approach and want to use DataAnnotation (or any other way, i.e. Fluent API) to specify allowed format of text. I.e. I want to use RegularExpressions to validate it etc. I want to do it on Data layer (not on client), I want to see constraints in SQL.
Is it possible?
If you are asking if you can see the regular expression constraint show up on the SQL Server side, the answer is no, SQL Server doesn't support Regex. If you want the EF engine to run validation for a DataAnnotation for a regular expression before it hits the database, then yes you can.
[System.ComponentModel.DataAnnotations.RegularExpression]
I need some architectural advice. I'm more into java, but trying to get up to speed with Ruby-on-rails. In the app I am building I need a convenient place to store some dictionary values that will be later used in various places of the application. These will be usually key value pairs - e.g. list of values to be used in select list.
The main objective is to keep this logic in one place of the application.
I am considering following options:
Store values in the database - i'm kind of reluctant from that, as values won't change very often.
Put all of the values in one class. In JAVA I'd have some static properties in one class holding this values (e.g. call Utils.getStates() will return list of states). How to do it ruby way?
Have some .yml file with values - read from the values. How to do it? I guess I have to parse the file in the initializer, but is there any tutorial how to do it?
Precise example? Let's say that have a model that have a field called "Type". Type can be: ['Type A', 'Type B', 'Type C'...]. And of course, for each type I want to have key and value.
I'd appreciate some suggestions about how you solve this problem in your apps.
Thanks,
Maciek
How often does the list change? Is it acceptable to have developers involved each time a value changes (updating code, re-deploying the app)? If the answer is no then store the values in a database.
Is the list of values reuseable? Then a gem or a yaml file with an initializer might be a good choice.
Is it just a small list and does not change often? Then you might want to consider a constant.
I think in Rails any data that would change at runtime and needs to be persisted, would normally be stored in the database. I think that would be the "rails way". You could save the data to yaml or json file, but that would not follow the normal flow of the MVC pattern that is so common in rails
Newbie here, wrapping my head 'round this stuff!
I'd like to use the hex number as my url (external identifier) and keep the uuid within the database for a ruby on rails application. Is this even possible?
Thanks a bunch
Many people advise you against it but, yes, it is possible. It will need some code for it, and the solution depends on which version of Rails you use and what you use for the database, which is why I'm going to answer in a generic way.
You will want to have two different fields for the model: one for the external hex representation and another one for a separate UUID. Then, you can use the hex string to find instances in your controller actions, for example.
Please take a look at the following (they don't seem to have the two fields but will point you to the right direction anyway):
Problems setting a custom primary key in a Rails 4 migration
Change Primary Key Issue Rails 4.0
http://www.speakingcode.com/2013/12/07/gracefully-using-custom-primary-keys-in-rails-4-routes-controllers-models-associations-and-migrations/
And a longer post of a similar thing to do: http://ruby-journal.com/how-to-override-default-primary-key-id-in-rails/
Also, the FriendlyId gem might do what you want.
LuaSQL, which seems to be the canonical library for most SQL database systems in Lua, doesn't seem to have any facilities for quoting/escaping values in queries. I'm writing an application that uses SQLite as a backend, and I'd love to use an interface like the one specified by Python's DB-API:
c.execute('select * from stocks where symbol=?', t)
but I'd even settle for something even dumber, like:
conn:execute("select * from stocks where symbol=" + luasql.sqlite.quote(t))
Are there any other Lua libraries that support quoting for SQLite? (LuaSQLite3 doesn't seem to.) Or am I missing something about LuaSQL? I'm worried about rolling my own solution (with regexes or something) and getting it wrong. Should I just write a wrapper for sqlite3_snprintf?
I haven't looked at LuaSQL in a while but last time I checked it didn't support it. I use Lua-Sqlite3.
require("sqlite3")
db = sqlite3.open_memory()
db:exec[[ CREATE TABLE tbl( first_name TEXT, last_name TEXT ); ]]
stmt = db:prepare[[ INSERT INTO tbl(first_name, last_name) VALUES(:first_name, :last_name) ]]
stmt:bind({first_name="hawkeye", last_name="pierce"}):exec()
stmt:bind({first_name="henry", last_name="blake"}):exec()
for r in db:rows("SELECT * FROM tbl") do
print(r.first_name,r.last_name)
end
LuaSQLite3 as well an any other low level binding to SQLite offers prepared statements with variable parameters; these use methods to bind values to the statement parameters. Since SQLite does not interpret the binding values, there is simply no possibility of an SQL injection. This is by far the safest (and best performing) approach.
uroc shows an example of using the bind methods with prepared statements.
By the way in Lua SQL there is an undocumented escape function for the sqlite3 driver in conn:escape where conn is a connection variable.
For example with the code
print ("con:escape works. test'test = "..con:escape("test'test"))
the result is:
con:escape works. test'test = test''test
I actually tried that to see what it'd do. Apparently there is also such a function for their postgres driver too. I found this by looking at the tests they had.
Hope this helps.
I am using propel orm as database abstraction layer.
I want to define different schema.xml files for diffenrent modules.
E.g. user.xml.schema for handling users and roles, or app.schema.xml for the application model.
My problem is, I want to reference to the user table of the user.schema.xml. I can handle this by the tag, but I want to use another table prefix for the user schema tables.
Running propel-gen creates two sql files (one for the user.schema and one for the app.schema), but the problem is, that the user-schema tables are generated twice. First with correct table prefix of the user schema file and the others with the prefix of the app schema file.
The foreign key also references to the wrong tables (that from the table with the prefix of the app.schema.xml).
I do not know any way to prevent this behaviour.
Any hints?
You cannot add different table prefixes to one database, and there is no way to add a tablePrefix attribute on the table tag. Let me explain a bit more, I know, you can specify a tablePrefix per XML schema even for a same database, but it leads to errors if you try to add relationship.. I don't know whether it's a bug or not, AFAIK the tablePrefix should be defined at the table level… Without relationship, you'll get a clean SQL file (or two if you don't set the propel.packageObjectModel build property to true.
You can read: http://www.propelorm.org/reference/schema.html. So I think, it's not possible to do what you would like to do unfortunately.
BTW, what you want to achieve is called multi component data model in the Propel doc.